Responsible innovation in biotechnology: Stakeholder attitudes and implications for research policy - Elementa: Science ...

Page created by Ernest Swanson
 
CONTINUE READING
Responsible innovation in biotechnology: Stakeholder attitudes and implications for research policy - Elementa: Science ...
Roberts, P, et al. 2020. Responsible innovation in biotechnology:
                                                                      Stakeholder attitudes and implications for research policy. Elem
                                                                      Sci Anth, 8: 47. DOI: https://doi.org/10.1525/elementa.446

RESEARCH ARTICLE

Responsible innovation in biotechnology: Stakeholder
attitudes and implications for research policy
Pat Roberts*,†, Joseph Herkert† and Jennifer Kuzma*,†

This article explores attitudes of stakeholders involved in biotechnology towards the Responsible
­Innovation (RI) framework. As a framework for governance, RI has received increasing scholarly attention
 but has yet to be successfully integrated into U.S. research and innovation policy. Using a mixed methods
 approach, we analyzed the attitudes of different biotechnology stakeholders, particularly those working
 in areas related to genetically modified organisms (GMOs) in agriculture and the environment, towards the
 principles and practices of RI. Homogenous focus groups (organized by stakeholder affiliation) and pre- and
 post-focus group surveys were used to measure attitudes towards RI. We designed the survey ­questions
 according to the Advocacy Coalition Framework (ACF) and examined the agreement of stakeholders
 with policy core beliefs (general principles of RI) and secondary beliefs (implementation practices of RI).
 Although all stakeholder groups had neutral to positive attitudes towards RI general principles, we found
 significant differences in their reactions to the scholarly definitions of RI and in their attitudes towards
 practices to implement RI. In comparison to government and advocacy groups, stakeholders promoting
 biotechnology innovations–industry, trade organizations, and academics–had more negative reactions to
 social science definitions of RI and to RI practices that relinquish control to people outside of technol-
 ogy development pipelines. Qualitative analysis of focus-groups revealed barriers for implementing RI
 practices. For example, innovators were cynical about including external voices in innovation pathways due
 to inflexible funding programs and were concerned about potential delays to innovation given the highly
 competitive environments for financing and patents. In order to help address these tensions, we call for
 the co-design of RI practices between biotechnology innovators and other stakeholders. The opening-up
 of biotechnology innovation to RI practices of anticipation, inclusion, responsiveness and reflexivity will
 likely be important for future, public legitimacy of emerging genetic engineering applications such as gene
 editing and gene drives.

Keywords: Responsible innovation; Biotechnology; Governance; GMOs

1. Introduction                                                      public input. While RI has gained traction in the European
Co-production emphasizes the way in which the technical              Union (EU) and UK in national policies (e.g. Macnaghten
and social are intertwined and mutually reinforcing; that            et al. 2016), the U.S. has yet to embrace it in research and
is, science and innovation do not exist in an autonomous             innovation systems.
sphere shielded from the social context in which they                  An area that could benefit from an integration of RI
occur (Jasanoff, 2005). Social construction theories have            is biotechnology. In particular, agricultural biotechnol-
spurred interest in the ways in which societal concerns,             ogy, including the development of GMOs in agricultural
values, and diverse points of view can be incorporated               and environmental systems, has been fraught with con-
into innovation pathways. Responsible innovation (RI) has            troversy. While most RI work has been done with newly
emerged as a framework to meet this challenge (Stilgoe               emerging technologies like geoengineering (Stilgoe 2015),
et al. 2013; Owen et al. 2013). RI brings science and tech-          agricultural biotechnology is a more mature area. It pro-
nological innovation into line with the democratic prin-             vides an interesting case in which to see the potential of
ciples of transparency, accountability, and meaningful               RI to shape a technology that already might be locked into
                                                                     certain paths in society (Collingridge 1980; Macnaghten
* School of Public and International Affairs, North Carolina State
                                                                     2016). The area of agricultural biotechnology, and more
  University, NC, US                                                 broadly, biotechnology applied to agriculture and the
†
    Genetic Engineering and Society Center, North Carolina State     environment, seems recalcitrant to the integration of
    University, NC, US                                               principles and processes associated with RI. For example,
Corresponding author: Jennifer Kuzma (jkuzma@ncsu.edu)               Marris (2015) found, through her embeddedness in UK
Responsible innovation in biotechnology: Stakeholder attitudes and implications for research policy - Elementa: Science ...
Art. 47, page 2 of 22                                                   Roberts et al: Responsible innovation in biotechnology

biotech communities, that synthetic biologists are hesi-      2. Background and framing of the analysis
tant to open up innovation processes to public engage-        2.1. Responsible innovation
ment due to potential public backlash, and they frame         RI has gained growing traction among those involved in
“responsibility” as a way to shut down conflict and garner    fields of science policy, science technology and society,
public acceptability. With decades of experience embed-       and the social studies of science. In the U.S., RI emerges
ded in US agricultural and environmental biotechnology        from previous scholarship including anticipatory govern-
communities, our observations are similar.1                   ance (Guston 2014) and midstream modulation (Fisher
   At the same time, even the most stalwart supporters        et al. 2006). Perhaps the closest that the U.S. has come
and developers of biotechnology recognize a need to           to incorporating RI into federal policy has been in refer-
change their approach to governance, recalling the mis-       ence to “responsible development” in the 21st Century
takes made from the 1st generation of genetically modified    Nanotechnology Research and Development Act (Public
organisms (GMOs) in agriculture. They are now seeking to      Law 108–153). In contrast, in the EU, principles of RI or
avoid these mistakes with emerging GM methods such as         its companion Responsible Research and Innovation (RRI)
gene editing and gene drives (Jordan et al. 2017; Kaebnick    have been incorporated into several supra-national and
et al. 2016; Kuzma et al. 2016; Kuzma et al. 2018). In this   national funding programs for over a decade. Recently, the
article, we explore whether and how RI might be accepted      adoption of RI goals, terminology, and ethos have been
into U.S. biotechnology innovation. We investigate the        incorporated into the Europeans Union’s Programme for
attitudes of diverse stakeholders working in areas related    Research and Innovation Horizon 2020 and Norway’s
to agricultural and environmental biotechnology to see        BIOTEK2021 program (von Schomberg 2013).
whether they agree with the general principles of RI as          One of the earliest definitions of RI describes it as
articulated by Stilgoe et al. (2013) and ways to implement
these principles.                                                an interactive process by which societal actors and
   Others have studied RI for biotechnology using global         innovators become mutually responsive to each
and developing-world case studies and by taking an               other with a view on the (ethical) acceptability, sus-
observational and anthropological approach (reviewed in          tainability and societal desirability of the innovation
Macnaghten 2016). They have also written on the practical        process and its marketable products (in order to allow
projects in the UK to integrate RI into synthetic biology        a proper embedding of scientific and technological
research and innovation (Macnaghten et al. 2016). Here           advances in our society) (von Schomberg, 2011).
we focus on the United States, in which biotechnology
(especially GM crops) has had a high level of placement in    This moves the obligations of scientists and innovators
society over several decades and may be locked in to path-    beyond the typical research integrity (e.g. animal care
ways of innovation that are recalcitrant to change. We also   ethics, anti-plagiarism, anti-falsification) to ensuring that
differentiate our work by using social science methods        ­science is “for, with, and in” society (Owen et al. 2013).
of group interviews (focus groups) and survey method-          Four key tenets of RI have been developed in the most
ologies to hear from biotechnology innovators2 and other       widely cited paper in this area: anticipation, reflexivity,
stakeholders from government, trade organizations, and         inclusion and responsiveness (Stilgoe et al. 2013). In this
consumer or environmental advocacy groups.                     paper, we use these four elements to structure our study
   We address the following questions: what do different       of stakeholder attitudes and describe each below:
stakeholders in U.S. biotechnology think about existing
academic and scholarly conceptions of RI?; where do they        • Anticipation has been adapted from concepts such
find the elements of RI problematic for or complementary          as anticipatory governance (Guston & Karinen, 2010;
to their experiences with U.S. biotechnology innovation           Guston, 2014) and midstream modulation (Fisher
systems?; and finally, what are the key societal, institu-        et al. 2006; Schuurbiers, 2011), but adds a forward-
tional, or attitudinal barriers to implementing policies          looking dimension where potential consequences are
and programs relating to RI?                                      explored and analyzed before technologies are fully
   Ultimately, this paper seeks to examine attitudes              developed.
towards RI among U.S. biotechnology stakeholders in             • Reflexivity moves governance of science and technol-
order to identify opportunities and challenges to imple-          ogy away from solely a risk-based approach to one
menting new governance policies and programs that could           that encompasses reflection on the underlying goals,
help incorporate public desires and concerns into research        motivations, limits of knowledge, assumptions, and
and innovation in biotechnology. To our knowledge, we             alternative framings of problems. It goes beyond pro-
report on the first project to apply mixed-methods (sur-          fessional self-critique of scientists and extends to re-
veys and focus group interviews) to compare U.S. biotech-         search funders, regulators, and the other institutions
nology stakeholder attitudes towards RI. Before moving            involved in technology development and governance
to our research approach and results, we summarize the            (Wynne 2006).
RI framework and related studies upon which we draw,            • Inclusion involves citizens and publics, in addition
briefly introduce the U.S. biotechnology context, and             to stakeholders, in governance of research and inno-
situate our work within the policy process theory of the          vation (von Schomberg, 2013; Stilgoe et al. 2013). It
Advocacy Coalition Framework (ACF).                               opens up the processes of reflexivity and anticipation
Responsible innovation in biotechnology: Stakeholder attitudes and implications for research policy - Elementa: Science ...
Roberts et al: Responsible innovation in biotechnology                                                    Art. 47, page 3 of 22

    to voices beyond those of subject-matter experts, mov-      also embedded themselves in a laboratory using a Socio-
    ing towards bottom-up policy making. It is built on         Technical Integration Research (STIR) protocol (Fisher &
    the notion that science and research are public goods       Schuurbiers 2013).
    (Bardone & Lind, 2016), and therefore discourse about          Several other projects have explored innovator attitudes
    them should be democratized under a Habermassian            towards RI outside of the U.S., such as in the EU, Canada,
    ideal of political legitimacy (Held, 1980).                 and UK. The EU’s SMART-MAP project convened research-
  • Responsiveness in turn takes the “what if” questions        ers in academe and industry, social scientists, and civil
    from anticipation, the discursive value of inclusive        society organizations to develop tools for incorporating RI
    engagement, and the collective process of reflexiv-         into innovation pathways (EU SMART-MAP 2017). Marris
    ity to influence the path of innovation (Bardone &          (2015) reflected on her experiences working within the UK
    Lind, 2016; Owen et al., 2013). It requires a capacity to   synthetic biology community to better understand natu-
    change shape or direction of innovation in response         ral scientists’ and engineer’s conceptions of RI. Hartley
    to stakeholder and public values, as well as new or         et al. (2017) investigate a case study of academic research-
    changing information or circumstances.                      ers working in multiple STEM fields at one UK ­university
                                                                through face-to-face interviews about RI. Carrier and
The RI framework is based upon the normative foundation         Gartzlaff (2019) interviewed researchers and research
that the goals of science, research, and innovation should      executives across the EU in a variety of technological fields
be guided by social desirability and legitimized through        across social and natural sciences, humanities, and engi-
open discourse. RI can also be characterized as a shift from    neering to study their understandings of RI. One study in
focusing on products and risk-based outcomes, to the pro-       Canada interviewed those who design, develop and com-
cesses of research and innovation (Jasanoff, 2005). Groves      mercialize health innovations about practices of RI (Rivard
(2011) argues in the context of medical technologies that       & Lehoux 2020). We reflect on these studies as they relate
RI involves a care for the future that goes beyond “bet-        to our results and conclusions in Sections 4 and 5.
ter rules”. Instead, RI requires changes to practices, insti-      In light of this previous work, our project is unique in
tutions, and the deep participation of publics in shaping       the following ways: 1) it draws upon a large sample to
innovation.                                                     compare meanings of RI across several stakeholder groups
   It is beyond the scope of the paper to reargue why RI        (over 100 participants from industry, trade organizations,
is normatively and substantively justified. For this pur-       consumer or environmental advocacy groups, academe,
pose, we point the reader to a body of previous scholarly       and government); 2) it focuses on the United States and
work (e.g. Wynne 2006; Jasasnoff 2007; Groves 2011;             on non-medical sectors of biotechnology; and 3) it uses
von Schomberg 2011; Owen et al. 2013; Stilgoe et al.            a mixed-method approach with both quantitative surveys
2013). Rather, we set out to understand stakeholder atti-       and qualitative focus-group data. Our study contributes to
tudes towards the predominant scholarly vision of RI as         the literature in other ways as well. First, we develop and
described in Stilgoe et al. (2013) and to identify barriers     implement a quantitative survey instrument that uses the
that exist for implementing RI within the U.S. biotechnol-      definitions of RI from a highly cited RI paper (Stilgoe et al.
ogy community. We also situate the study of these atti-         2013) and that is inspired by ACF theory (see Section 2.3).
tudes within the Advocacy Coalition Framework (ACF)             We also focus on a more specific and controversial field of
(Section 2.3 below), which is a theoretical framework           biotechnology, the agricultural and environmental release
within the policy process literature that examines how          of GMOs. Finally, through focus groups, we identify bar-
individual and group beliefs relate to the formation of         riers to RI across multiple types of institutions (NGOs,
coalitions in policy arenas (Jenkins-Smith et al. 2014).        government, industry, academe, and trade organizations),
   Despite a large body of scholarly work on RI, few pro-       which all influence biotechnology innovation.
jects have taken a systematic approach to finding out what
stakeholders operating within innovation systems think          2.2. Biotechnology innovation and regulation
about RI, and even fewer have considered stakeholder            Biotechnology has, and continues to be, the subject of much
attitudes within U.S. biotechnology innovation systems.         discussion pertaining to questions concerning the integra-
Our is the first to use both qualitative and quantitative       tion of the social and the technical. Genetic engineering
approaches to assess attitudes of stakeholders from differ-     (GE) technologies are rapidly changing food, medical, pub-
ent sectors involved in U.S. biotechnology innovation. Two      lic health, and industrial systems (Chakradhar, 2015; Bailey,
previous studies were found that focused on U.S. academic       2015; Kuzma, 2016). Products of genetic engineering are
researchers at single institutions. One study interviewed       being released into the environment and agricultural sys-
U.S. innovators coming from a biotechnology institution         tems at an increasing rate, and despite the public contro-
within a university (Doezema & Guston 2018). However, in        versies, GE products have been a market success.
this study, RRI was framed a bit differently than in Stilgoe       GE product and GMO releases in the United States
et al. (2013). Core RI elements fell into categories of eth-    are regulated under the rubric of the 1986 Coordinated
ics, science education, open science, societal engagement,      Framework for the Regulation of Biotechnology (CFRB)
gender equality, and diversity. In another study, Glerup        (Fed. Registry 51, 23302; 1986). The CFRB evolved out of
et al. (2017) interviewed 11 U.S. academic scientists           a neoliberal political context in the mid-1980s that was
working in synthetic biology and nanotechnology and             predominantly occupied with falling behind in global
Responsible innovation in biotechnology: Stakeholder attitudes and implications for research policy - Elementa: Science ...
Art. 47, page 4 of 22                                                    Roberts et al: Responsible innovation in biotechnology

competitiveness. Economic growth through science and           Framework (ACF) considers the policy process as occurring
technology, including through newly emerging biotech-          within policy subsystems, like biotechnology innovation,
nologies, was seen as a panacea for this concern. During       which are bound by topic or scope and contain a variety
this time, the Reagan administration brought with it the       of actors seeking to exert influence on policy outcomes
philosophy of ‘new federalism’ which focused on disman-        (Jenkins-Smith et al. 2014). The framework suggests that
tling the bureaucracy, devolving authority to the states,      actors form advocacy coalitions based on shared norma-
and deregulating in areas of public health, occupational       tive beliefs and strategies for achieving desired outcomes
safety and the environment (Steinzor 1996). At the same        (Jenkins-Smith et al. 2014). Within the ACF, actors’ belief
time, academic and industry relationships were forged to       systems are described by three levels that move from the
increase the opportunities and incentives for the trans-       deep cultural beliefs to more specific views about policies
fer of technology outside of the lab to the marketplace.       associated with the policy subsystem and ways to imple-
Under the Bayh-Dole Act of 1986, universities, scientists,     ment those policies. “Deep core beliefs” are the most fun-
and small businesses receiving federal funds were allowed      damental to a person’s cultural identity and are unlikely
to retain the ownership of patent rights that came about       to be changed. “Policy core beliefs” are fundamental
from government funded projects, a shift towards a free-       views about principles under which the policy subsystem
market approach that would give shape to US science            should operate, and they are often shared within stake-
policy for decades to come (Markel 2013).                      holder groups or coalitions that work closely together. The
   The solidification of university-industry partnerships      third level is composed of “secondary beliefs” that reflect
has led to what some scholars have termed “academic            how policies should be implemented to achieve desired
capitalism”, which is the reprioritization of universities     outcomes. These are usually the most malleable beliefs,
towards pursuing knowledge with the potential of rev-          and frequently change in response to new information
enue generation (Rhoades & Slaughter 2004). Similarly,         (Henry, Lubell, & McCoy 2011).
the Stevenson-Wydler Technology Innovation Act of 1980            In this paper, we test “policy core beliefs” (aka RI prin-
allowed federal laboratories to actively engage in technol-    ciples) and “secondary beliefs” (aka RI implementation),
ogy transfer of their work (Slaughter and Rhoades 2004).       but not the deepest core beliefs in the ACF. We designed
The modern academe-government-Industry relationship,           survey questions to reflect the four core principles of RI—
with strengthened university-Industry partnerships, was        anticipation, reflexivity, responsiveness, and inclusion as
thus born. Government policies emerged that were specif-       described in Stilgoe et al. (2013)—at the ACF’s level of gen-
ically geared towards incentivizing universities to engage     eral “policy core beliefs” (Table 1). Our policy core belief
in commercialization with industry (Meyer 2003).               questions were specifically designed to capture the four
   The close-knit relationship between academe, industry,      RI principles at the level of ACF policy core beliefs accord-
and government prevail today in science and technology         ing to Jenkins-Smith (2014). We also chose one possible
development which is still seen as the primary force for       way to implement those four RI principles from the alter-
economic growth. Responsible Innovation may disrupt            natives discussed in Stilgoe et al. (2013) for our survey
these relationships however, with its call for opening the     questions. These questions were designed to represent
university-government-industry triangle to public input        ACF “secondary beliefs” and RI implementation practices
and responsiveness. Thus, it was important for us to           (Table 1).
explore the attitudes of these stakeholder groups towards         Our goal was to see whether the different biotechnol-
RI to identify potential barriers. We expected university,     ogy stakeholder groups agreed on RI principles (policy
industry, and government groups to hold similar attitudes      core beliefs) and/or implementation (secondary beliefs).
towards principles and practices of RI. We also thought the    ACF posits that if coalitions agree on policy principles
views of trade organizations would align with this triad,      (policy core beliefs), they are more likely to form closer
as they often represent the viewpoints of biotechnology        and stable working relationships to advocate for policies
developers (e.g. bioindustry associations); scientific soci-   and programs. This is because disagreements over second-
eties; or biotech user communities (e.g. farming associa-      ary beliefs (RI implementation) among coalitions or stake-
tions whose members grow GMO crops). In contrast, we           holder groups are easier to manage, as these beliefs can
suspected that the attitudes of consumer and environmen-       change over time, under different circumstances, or upon
tal advocacy groups towards RI would differ. These groups      new information. Thus, an even deeper motivation of our
often strive to be included in the innovation system and       research was to explore the potential for future advocacy
have an adversarial relationship with biotech developers. In   coalition formation that transcends or changes current
fact, legal cases and media attention sparked by the activ-    entrenched U.S. coalitions of “pro-biotech” (largely indus-
ity of advocacy groups have prompted change in the CFRB        try and academic developers) versus “anti-biotech” (largely
over time at key junctures (e.g. see Kuzma 2020). These        advocacy groups).
hypotheses and ones from the ACF, as described below,              We also examined whether attitudes towards RI princi-
informed the structure of our research study, the survey       ples (ACF policy core beliefs) or implementation (ACF sec-
questions, and the coding of the focus-group results.          ondary beliefs) shifted after exposure to more information
                                                               about the predominant scholarly vision for RI (Stilgoe et al.
2.3. Advocacy coalition framework                              2013). As described in more detail in Section 3, scholarly
Given the complex, interactive, and temporal nature of         definitions for the four tenets of RI as derived from Stilgoe
the policy process, various theories have been developed       et al. (2013) (Table A3 Appendix) were presented to our
to make sense of policymaking. The Advocacy Coalition          participants in focus groups after they took a pre-survey
Roberts et al: Responsible innovation in biotechnology                                                       Art. 47, page 5 of 22

Table 1: Core elements of RI and survey questions. DOI: https://doi.org/10.1525/elementa.446.t1

 RI Dimension       Survey Question RI policy principles              Survey Question RI policy implementation
                    (ACF policy core belief)                          (ACF secondary beliefs)
 Inclusion          Maximizing public participation leads to better Innovators should consult with consumer and environmen-
                    biotechnology policy.                           tal advocacy groups during R&D in biotech.
 Reflexivity        Reflecting on the underlying purposes, motiva- Social scientists, environmental and health risk analysis and
                    tions, and uncertainties that surround biotech- ethicists should be involved from the early stages of biotech
                    nology products is important.                   innovation.
 Anticipation       Considering potential environmental and social There should be a standard of at least 10% of public funding
                    implications of biotechnology products is impor- for research in biotechnology that goes to environmental,
                    tant in the planning stages of research.         social, legal, and ethical implications research.
 Responsiveness The innovation process should respond to changes The innovation process should respond to changes in public
                in public attitudes or values.                   attitudes or values even if this means delaying, modifying or
                                                                 terminating the project.

but before they took the post-survey. The same questions          (related to non-medical biotechnology industries), or con-
for ACF policy core beliefs and ACF secondary beliefs were        sumer/environmental groups (also related to non-medi-
asked in the pre- and post-tests to detect possible changes       cal biotechnologies). Academic and industry participants
in attitudes towards RI principles and implementation             were chosen based on their conduct of “use-inspired”
upon this new information. We analyzed the portion of             or applied research related to biotechnology products
the focus group discussions right after these definitions         applied to environmental or agricultural applications.
were presented to explore what the initial reactions of the       The rationale behind choosing to split non-governmental
stakeholders were to scholarly visions for RI. These results      organizations (NGOs) between trade organizations and
are discussed in Section 4.2.                                     consumer/environmental groups was that trade organiza-
                                                                  tions can wield a significant amount of pressure on the
3. Methodology                                                    political system through lobbying on behalf of private
An approach mixing qualitative and quantitative data was          interests in American politics. In contrast, consumer and
employed to assess stakeholder attitudes towards RI, as           environmental groups yield pressure outside the system;
well as how they reacted to the academic conceptualiza-           for example through the media or courts, which can
tion of the RI framework, particularly the four elements          ­significantly impact the dynamics of policy change in bio-
of RI from Stilgoe et al. (2013). Survey items were used to        technology oversight (Kuzma 2013; Kuzma 2020). Govern-
capture stakeholder attitudes towards the principles and           ment representatives were from policy-making or analysis
practices of RI, while focus groups allowed for a more in-         areas related to biotechnology in food, agriculture, health
depth exploration of how representatives from each stake-          and the environment.
holder group reacted to the scholarly definition of RI and            Focus group participants were recruited from a sample
the barriers they saw to implementing RI. Following the            of stakeholders in the greater Raleigh-Durham-Chapel Hill
logic of Kelle and Erzberger (2004), this mixed-methods            area (Research Triangle, NC) whose work related to the
approach was used to reveal a fuller and richer picture of         application of biotechnology in food, agriculture, and the
how the different sectors approached the concepts of RI.           environment. This area hosts a rich diversity of organiza-
   The focus group also served as an intervention:                 tions working in the biosciences. Although chosen in part
1) ­agricultural and environmental biotechnology stake-            because of its proximity to the research team for conveni-
holders were surveyed prior to the focus groups (pre-              ence, The Research Triangle region appears as the 2nd to
survey); 2) during the focus groups they were exposed to           10th most active biosciences region in the U.S. across mul-
definitions of the four elements of RI (Stilgoe et al. 2013)       tiple reports and metrics. In 2015, it ranked only second to
and asked to discuss their reactions to them in the con-           the Boston area as a life sciences hotbed according to the
text of biotechnology innovation (intervention); and then          JLL Life Sciences Outlook Report (Rose 2015). Although
3) surveyed with the same questions (post-survey). We              our participants are not a nationally U.S. representative
used the intervention to test hypotheses regarding how             population, they do reflect a representation of one of the
­different stakeholders view RI as it is envisioned in the         nation’s most active biotech regions.
 academic and policy literature and to identify challenges            For participant recruitment, we first developed a spread-
 and opportunities different biotech stakeholders perceive         sheet of contacts from the Genetic Engineering and Society
 to implementing RI as defined by RI scholars.                     (GES) Center database and listserve and from people with
                                                                   whom we have interacted. We then asked key contacts and
3.1. Study participants                                            collaborators to provide more names through a snowball
The participants chosen for the study were selected based          sampling strategy. For stakeholder groups for which our
on the demographic criteria of profession, which in this           contacts were limited, we searched website databases for
case means that they were employed in a capacity related           organizations in the area whose work relates to biotech-
to agricultural or environmental biotechnology innova-             nology (particularly its use in agriculture or the environ-
tion: academe, industry, government, trade organizations           ment). We ultimately developed a spreadsheet of over
Art. 47, page 6 of 22                                                    Roberts et al: Responsible innovation in biotechnology

700 possible participants and recruited 104 participants       distinct positions that are associated with biotechnology
(Table A1, Appendix).                                          products. However, they are a construction of a conversa-
   Participants were first recruited by an introductory        tion that would not necessarily happen naturally as mem-
all-group email explaining the project and containing a        bers are encouraged to react to each others’ viewpoints,
form outlining their rights as participants (IRB protocol      and thus group effects of dominant personalities or
was approved by NC State University as “exempt”). Non-         ­hierarchies need to be considered during the moderation
responsive invitations were followed up with a second           (Kitzinger 1995).
email reminder, and then, if necessary, targeted phone             Moderators of our focus groups were graduate students
calls were made to fill stakeholder groups for which par-       in the biological and social sciences who were trained
ticipants were lacking.                                         through a rigorous four-day workshop on RI and focus
   The demographic composition of the groups in Table A1        group methodologies, including ways to minimizing
(appendix) shows a largely white, middle-aged population        over-bearing or distractive participants and foster positive
of biotechnology innovators and policy makers in aca-           discussion (Herkert et al. 2017). Limson (2018) also used
deme, government, industry, and trade organizations.            South African bioscience graduate students to engage
This is generally what we observe in our interactions with      with publics and found that they gained a greater sense
stakeholders in agricultural biotechnologies as well. The       of motivation to do research that benefits the public
exception was that our consumer-environmental organi-           which resonated with RRI learning outcomes. Our project
zation participant group tended to include more female          engaged graduate students to interact with stakeholders
participants and had the youngest mean age of about 43.         about RI through focus groups and found similar results
This reflects our observations from our engagement activ-       with regard to RI outcomes, which are reported elsewhere
ities with these stakeholders and supports the literature       (Herkert et al. 2017).
that females tend to place more importance on the envi-            The focus groups were structured in a way to allow
ronmental and health risks of emerging technologies than        for replicability in terms of the process and structure
males (Finucane et al. 2000). Overall, the demographics         (Bohnsack 2004). All the focus groups were conducted
reflected our observations of the field as experienced in       under the same protocol, which consisted of a discussion
the Research Triangle, although we do not make the claim        moderated by a trained moderator who used the same set
that they are representative of U.S. biotechnology sectors.     of structured questions and followed in the same order.
                                                                However, moderators were engaged in topical steering
3.2. Focus group research                                       rather than being rigid in moving from topic to topic.
The use of homogenous focus group is part of a methodo-         Topical steering avoids what Merton et al. (1987) refer to
logical foundation originating in the tradition of group        as the fallacy of fixed questions, which can cut off fruit-
discussion (Morgan 1997), where discussion groups are           ful discussion. Moderators were thus given the ability to
taken as examples of the broader communities although           channel the discussion in the direction needed to obtain
they may not be statistically representative. Groups are set    relevant data without forcing the group to respond in
up to have commonalities in professional aspects related        specific ways. As Morgan (1997) notes, avoiding rigid dis-
to the topics under discussion. We hosted three focus           cussion also keeps the moderator from trying to exert too
groups per sector (with the exception of 2 groups for con-      much influence to produce ideal data.
sumer-environmental NGOs), and on a sector basis, the              The questions that moderators used in the focus groups
discussions yielded similar themes across these replicates.     are listed fully in the appendix (Table A2, Appendix).
Therefore, although our focus groups are not meant to           Discussion questions proceeded from general ques-
contain a demographically representative sample of U.S.         tions about responsibility to questions surrounding the
biotechnology sectors, the themes from the discussions          formal RI framework as proposed by scholars (Table A3,
seem representative of the sectors.                             Appendix). Two note takers were present to record visual
   Homogenous focus groups (as opposed to heteroge-             cues and observe larger themes from the conversations.
neous groups typically utilized in market research) treat       In this paper, we report only on the discussions follow-
the discourse of the focus groups, and consequently the         ing question 7, which presented the academic definitions
textual data derived from such discourse, as reflective of      of the four elements of RI—anticipation, responsiveness,
the group (Bohnsack, 2004). However, discussions emerge         inclusion, and reflexivity as described by STilgoe et al.
in the group that may not emerge from individual inter-         (2013)—and asked for participants’ responses to this schol-
views, as members of the group build upon and react to          arly definition (Table A2 and A3 in Appendix).
others’ arguments or points. Focus groups encourage par-           All 14 focus groups (2 for consumer-environmental
ticipation by people hesitant to be interviewed on their        NGOs, 3 each for the 4 other stakeholder categories,) fol-
own or lack confidence that they have anything to con-          lowed the same set of procedures and were 90 to 105 min-
tribute (Kitzinger 1995). Homogenous focus groups also          utes in length. Focus groups contained between 4 to 12
capitalize on members’ shared experiences (Kitzinger            participants and were conducted in May 2016 and 2017.
1995). Employment Sector was selected as the character-         We detected no significant differences in content depend-
istic for dividing participants based upon the idea that        ing on group size within a sector, although we note that
policy values and stances on biotechnology are likely to        the consumer-environmental NGO focus groups were
vary by sector since at a broad level sectors tend to play      smaller in size. Recruitment for consumer-environmen-
different roles in the innovation process and advocate          tal NGO participants was more difficult in the Research
Roberts et al: Responsible innovation in biotechnology                                                      Art. 47, page 7 of 22

Triangle area as few groups in this region make biotech-        Specifically, we drew from Stilgoe et al.’s (2013) textual
nology policy a priority. To increase participation in this     descriptions of the four elements—anticipation, reflexiv-
area, we included sustainable agriculture groups, organic       ity, inclusion, and responsiveness–for RI principle state-
agricultural groups, and general environmental policy           ments (aka policy core beliefs according to ACF) and from
groups to help recruit participants.                            their “Indicative techniques and approaches” (Stilgoe
  We focus in this paper on what stakeholders thought           et al. 2013, p. 1573) for the questions about policy imple-
about scholarly definitions of RI and what barriers and         mentation (aka ACF secondary beliefs). In developing our
opportunities they saw for implementing RI practices and        ­questions, we made particular choices for RI implemen-
policies to achieve the four elements of RI (Stilgoe et al.      tation based on our understanding of the literature and
2013; Table A2 & A3). Thus, we report in this paper on the       experience with the field of RI.
qualitative discussions following the question:                     For example, in the case of anticipation, ethical, legal
                                                                 and societal implications (ELSI) research done upstream of
   Here are some key elements of RI from the perspec-            technology development is a cornerstone of this RI prin-
   tive of social scientists who study emerging tech-            ciple (Stilgoe et al. 2013), as well as its precursor, anticipa-
   nologies: We are not suggesting that these are right          tory governance (Guston & Karinen, 2010; Guston, 2014).
   or wrong, or recommending them. But we’d like to              Anticipation is a “systematic thinking of the many differ-
   know what you think about them. Take some time to             ent potential outcomes” from an emerging technology
   read examples of these elements: What do you think            (Felt 2018), which directly ties anticipation to the study
   of these definitions and elements of RI?                      of ethical, legal, and societal implications. For ELSI study,
                                                                 one must have resources early on in technology develop-
Time spent on this question averaged between 10 to 20            ment to do anticipation of downstream consequences.
minutes for each focus group. Other analyses and papers          Thus, funding ELSI work is an important component of
are in preparation and will report on stakeholder’s bot-         anticipation. RI scholars and some policy makers have
tom-up meanings of RI from the other focus group ques-           called for greater funding of upstream ELSI research as
tions (see Table A2, Appendix).                                  a component of anticipatory governance (Guston 2014)
   Iterative thematic coding using NVivo software was            and now RI (Felt 2018). Funding of ELSI shows commit-
used for qualitative analysis of the transcripts. Themes         ment to implement (through investment) the RI principle
and coding structure were designed by two of the three           of anticipation. Ten percent ELSI funding in a budget for
paper authors in a consensus and iterative coding process.       a national technology program seemed a realistic, but
The coding was conducted by one of the coauthors, and a          ambitious, goal for the U.S., as it increases the amount cur-
sample of the coding for each theme was evaluated by the         rently spent on ELSI by about 2 to 10 times over past U.S.
second author until agreement was achieved. We based             programs like the Human Genome Project and National
initial themes on the four elements of RI as described in        Nanotechnology Initiative (Bennett & Sarewitz 2006).
Stilgoe et al. (2013), read the transcripts and coded for           However, we cannot claim that the questions in
them, and then looked for additional themes and sub-             Table 1 are the only choices that could have been made for
themes to clarify the participants’ attitudes and identify       implementing anticipation and other RI principles. Thus,
barriers and challenges to implementing RI.                      our results are constrained by the RI principles in Stilgoe
                                                                 et al. (2013) and by our desire to use a reasonable set of
3.3. Quantitative survey methods and analysis                    questions among the many alternatives for RI that schol-
Participants in the study were subjected to a pre-focus          ars propose. We discuss these and other limitations of our
group survey and a post-focus group survey. We adminis-          approach in the final section of the paper (Section 5.1).
tered the same instrument before and after the focus group          Participant responses to these questions were measured
which included formal exposure to the academic defini-           using a 7-point Likert scale that ranged from strongly disa-
tions of RI. At the time of study design (early 2016), we did    gree to strongly agree. Questions were designed as some-
not find any developed survey instruments for assessing          what general open to interpretation so that priming prior
agreement with RI in the literature. Thus, we developed          to the focus groups would be minimal. In this case, the RI
four survey questions according to the description of RI         definitions presented in the focus groups would serve to
principles and practices in Stilgoe et al. (2013) (Table 1).     clarify the meaning of the four terms.
We justify a focus on the Stilgoe et al. (2013) in our survey       Participants’ written survey responses were kept con-
and focus groups paper as it is the most highly cited paper      fidential and separated from the participant’s name (by
when one searches for “responsible innovation” in Google         assigning them numbers) according to our approved IRB
Scholar (over 1236 citations with the next paper at 379 as       protocol. Pre-surveys were sent 1 week prior to the event
of April 24 2020). We tested the survey instrument prior         and filled out online prior to the focus group. Post surveys
to using it on the study participants using a test group         were completed immediately in-person after the focus
of biotechnology stakeholders who serve on our Center’s          group. Within the focus groups, participants were referred
advisory board committee. We revised the questions based         to by a number rather than name to increase confidential-
on their feedback on the survey.                                 ity of oral comments, although we could not avoid some
   The final questions in Table 1 were used for the par-         group members knowing each other by name in a profes-
ticipants to choose their level of disagreement or agree-        sional capacity. Participants were offered a $50 Amazon
ment with principles or practices (implementation) of RI.        gift card as an incentive following their completion of
Art. 47, page 8 of 22                                                     Roberts et al: Responsible innovation in biotechnology

the final, post survey (some stakeholders declined or           prior to the focus groups; 2) scholarly definitions of RI
requested a reduced amount based on their organization’s        presented at the focus groups in between the surveys
policies).                                                      may have had a polarizing effect on stakeholder attitudes
                                                                towards RI, particularly between industry and consumer-
3.4. One-way analysis of variance (ANOVA):                      environmental groups; 3) with the exception of i­ nclusion,
Stakeholder Differences Pre and Post-Test                       stakeholder groups involved in innovation (industry,
We expected, due to the features of the biotechnology           trade, and academe) disagreed more with the implemen-
innovation system discussed in Section 2.2, that industry,      tation of RI (secondary beliefs) than general principles
trade organizations, government, and academic stake-            of RI (policy core beliefs); 4) the greatest magnitude of
holders would align in their attitudes towards RI prior         disagreement occurred between industry and consumer-
to focus groups given the tight relationships and innova-       environmental advocacy groups with the RI principles of
tion goals in U.S. political systems. Our initial hypothesis    responsiveness and inclusion; 5) the general principle of
was that there would be no significant differences among        anticipation (policy core belief) showed little difference
these groups given the triple helix (with trade organiza-       among stakeholder groups, especially prior to the focus
tions representing industry as well). Given the conten-         groups; however, industry strongly opposed the way antic-
tiousness of the subject matter and history of the GMO          ipation was implemented (secondary belief) with greater
debates (Kuzma et al. 2009; Kuzma 2013), we expected            ELSI funding as the proposed policy tool; 6) government
significant differences in pre-test scores between con-         aligned more with consumer-environmental groups than
sumer-environmental groups and industry (and likely             other stakeholder groups and 7) industry, trade orgs, and
government, trade orgs, and academe as parts of the triple      academe generally aligned in their distaste for inclusion
helix). We used descriptive statistics to observe any differ-   and responsiveness. We describe the above survey results
ences among groups separately in the pre-test or post-test,     in more detail below and return to the focus groups and
and then tested for statistically significant differences by    identification of policy barriers in section 4.2.
measuring analysis of variance (ANOVA) between pairwise
groups. A one-way between-subjects design is commonly           4.1.1. Descriptive statistics for attitudes towards RI
used for non-experimental survey research where groups           Although our study population was relatively large for
are differentiated by demographic data (in this case sector      exploring expert and stakeholder attitudes in a particular
of employment) and are treated as independent variables          technology domain (n > 100), it was regionally limited to
(Meyers et al. 2013). Statistical significance was tested for    the North Carolina Research Triangle region and does not
by comparing the F test statistic. A Levene’s test for homo-     represent biotech stakeholders across the United States.
geneity of variance was also conducted as ANOVA requires         Descriptive statistics are used in studies with smaller num-
the variance between groups to be homogenous, although           bers of participants or those that are not representative
this may be relaxed in certain cases. The Levene’s test uses     of the entire population to explore patterns in the data.
an F-test to test the null hypothesis that the variance is       Thus, we chose to evaluate the data descriptively first,
equal across groups.                                             before doing significance testing (Sections 4.1.2 & 4.1.3
   As a one-way analysis test only computes a single p-value     below). In Figure 1, the mean scores of the participants’
indicating whether a difference in mean exists between           rankings of RI principles (policy core beliefs) and RI imple-
groups, it does not indicate which ones differ, so we added      mentation practices (secondary beliefs) are color-coded
the Tukey’s honest significance difference (HSD) test to         according to the level of agreement (green) or disagree-
perform a single-step multiple pairwise comparison to            ment (red) or neutral (white). Both the pre-test and post-
compare all possible pairs of means. The Tukey HSD test          test results are shown.
provides a more conservative adjusted p-value to account           In the pre-test, we found neutral to positive (4 to =7)
for family-wise error and is best for groups with different      ratings for RI principles (policy core beliefs) across all
sample sizes (such as our stakeholder groups, Table A1).         stakeholder groups. However, RI principles of inclu-
                                                                 sion and responsiveness were rated more neutrally; with
4. Results                                                       greater support for reflexivity and anticipation (Figure 1;
4.1. Survey results                                             Table A4, Appendix). Consumer organizations were most
Our findings suggest (Figures 1 and 2, Tables 2–4) that:        ­supportive of responsiveness, while industry, trade organ-
1) all stakeholder groups showed neutral or slightly posi-       izations, and academe expressed more neutral attitudes.
tive attitudes towards RI principles (policy core beliefs)       The RI principle of inclusion was most highly supported

Figure 1: Visualization of scores. Scale of Red-White-Green from 3.0 to 7.0 respectfully. Stakeholder groups in ascend-
  ing order from left to right for agreement with RI. Core = RI principles or ACF policy core believes; Implementation =
  RI implementations or ACF secondary beliefs. Pre = survey before focus groups; Post = same survey after focus groups.
  DOI: https://doi.org/10.1525/elementa.446.f1
Roberts et al: Responsible innovation in biotechnology                                                        Art. 47, page 9 of 22

Figure 2: Cumulative graphs of four elements of RI. Groups coded by I = industry, A = academe, T = trade organi-
  zations, G = government, C = consumer (and environmental) NGOs. Pre = pre-focus group, Post = post-focus group.
  Core = core principles of RI, Impl = implementation of RI element. The dashed line shows the cumulative distribution
  of the scores or percent which are above to the left and below to the right of the line. DOI: https://doi.org/10.1525/
  elementa.446.f2

by consumer-environmental organizations followed by               practices of RI (secondary beliefs) with the highest mean
government, then academe, and finally industry and trade          scores (Figure 1; Table A5). Conversely, industry ranked
organizations. RI principles of anticipation and reflexivity      each of the practices proposed to implement RI lower than
were favorably rated by all groups (Figure 1).                    other stakeholder groups. In particular, the implementa-
   In comparing the post-test with the pre-test for RI gen-       tion of anticipation was scored very low by industry. This
eral principles (policy core beliefs), we noticed that industry   is likely due to the specific way we operationalized antici-
was the only group whose mean score decreased across              pation, as 10% funding for ELSI research (Table 1). The
all four principles; and that consumer-environmental              results reveal that biotech innovators in this case do not
organizations were the only group that saw an increase            highly value the integration of social science and humani-
in mean scores across all four principles (Figure 1; Table        ties scholars in anticipatory governance if it involves fund-
A4 Appendix). This suggested to us that the focus group,          ing their work.
which included a presentation of the scholarly definition            For academe, post-test scores for implementation of
of RI (Section 3.2), was an intervention that seemed to           RI decreased; whereas for industry, they stayed largely
polarize the attitudes of industry and consumer-environ-          the same with only slight decreases in anticipation and
mental organizations. Interestingly, government scores            responsiveness. However, industry scores remained lower
increased for three of the four principles after the focus        than other groups. (Figure 1; Table A5). In Section 4.2,
groups: all but anticipation.                                     we use the focus group discussions that occurred after the
   When the core RI principles were operationalized               scholarly definition of RI was presented to examine why
into RI practices (secondary beliefs or RI implementation)        these innovators might be hesitant to endorse the ways
(Table 1), industry, trade organizations, and academic            we chose to implement RI in the survey questions.
participants—the developers of biotech products—were
less in favor of them (Figure 1, Table A5, Appendix) when         4.1.2. Significance of differences towards RI principles (ACF
compared to consumer-environmental and government                 policy core beliefs)
groups. This suggests to us that it is easier for developers      Through using descriptive statistics, we found ­differences
of biotechnology to accept RI in theory, at the level of ACF      of means in the pre-test scores among stakeholder groups
policy core beliefs, rather than in practice, at the level of     for the general principles of RI of up to 0.8 to 0.9 (­Figure 1).
ACF secondary beliefs. We note the caveat that our survey         However, the ANOVA results with a post-hoc Tukey HSD
questions presented a limited set of choices for imple-           test showed that none of the groups had statistically sig-
menting RI (discussed more in Section 5.1).                       nificantly different mean scores (Table A3, Appendix). This
   As with the general principles, consumer organiza-             is likely due to the high variation of scores within each
tions displayed the most favorable attitudes towards all          group. However, after exposure to the ­scholarly definition
Art. 47, page 10 of 22                                                        Roberts et al: Responsible innovation in biotechnology

of RI during the focus groups, in the post-test, industry          Overall, it is interesting that the presentation of the
scores for the general principles of RI (ACF core policy         scholarly definition of RI in the focus groups brought
beliefs) were statistically different in comparison to other     these differences to light in comparison to the pre-test.
stakeholder groups (Table 2).
   For the principle of inclusion, industry scores were          4.1.3. Significance of differences towards implementation of
statistically lower than all groups but trade organiza-          RI (ACF secondary beliefs)
tions. The pairwise comparison between academe and               In contrast to the general RI principles, pretest scores for RI
industry showed a mean difference of –1.32, between              implementation practices (ACF secondary beliefs) revealed
government and industry of –1.77, and consumer-envi-             significant differences across some groups (Table 3). These
ronmental organizations of –2.04 (Table 2). Because              were again seen between industry and all groups but trade
government, academe (mainly from public universities),           organizations. Interestingly, implementing anticipatory
and consumer groups represent the public sector, these           practices was more problematic to industry in compari-
groups may feel more of an obligation to include exter-          son to consumer (–2.88), academe (–1.69), and govern-
nal voices in innovation systems, at least in theory if not      ment (–1.73) groups. We also saw a significant difference
practice (Sections 4.1.3 & 4.1.4).                               between trade and consumer groups in implementing
   Industry differed across the board from consumer              anticipation (–1.61). That is, industry or trade o
                                                                                                                  ­ rganizations
groups for all three other principles, ranging from –1.11        disagreed more than these other groups with increasing
for reflexivity, to –1.40 for anticipation and –1.87 for         ELSI f­ unding as the way to implement anticipation.
responsiveness. We observed again that responsiveness               Industry and trade organizations also rated the imple-
and inclusion were the most negatively perceived by the          mentation of inclusion significantly lower than consumer-
key sector of innovation in biotechnology, the industry.         environmental groups (–1.66 and –1.59 respectively).
For both principles, according to the scholarly defini-          Industry rated reflexivity practices significantly lower
tion of RI, industry should share control of innovation by       than consumer-environmental groups as well (–1.95).
opening up the process to external voices (inclusion) and        Interestingly, there were no significant differences among
responding to their concerns (responsiveness). Industry          groups for the RI implementation of responsiveness, in
may not want to relinquish their power of the course of          contrast to the significant disagreements over the RI prin-
innovation (more on this in Section 4.2 and 5).                  ciple of responsiveness discussed above (Section 4.1.2).
   Consumer-environmental group scores were also sig-               In the post-test, the above differences in attitudes
nificantly higher than academic scores with regard to the        towards implementing RI remained and additional ones
RI principle of responsiveness (1.48). Although academics        arose (Table 4), again indicating the polarizing effect of
may be inclined to include the public, they may be less          the focus groups and presentation of the scholarly defini-
receptive to responding to public voices in research and         tion of RI. Industry again differed from consumer-environ-
innovation for reasons we discuss in section 4.2, such as        mental groups across the board on all 4 implementation
inabilities of these researchers to change directions due to     practices of RI and from government and academe on
inflexible contracts and grant funding.                          implementing anticipation. Trade organizations again

Table 2: RI principles after focus groups—significance of differences among stakeholder groups. Tukey mul-
  tiple comparisons of RI principles – Posttest means, 95% Family-wise confidence Interval: p-value (adjusted mean
  difference). DOI: https://doi.org/10.1525/elementa.446.t2

                  Group – Group                                         Principle
                                            Inclusion        Anticipation      Responsiveness       Reflexivity
                  Consumer – Academe           0.57 (0.71)      0.29 (1.01)        0.014* (1.48)       0.35 (0.60)
                  Government – Academe         0.88 (0.44)      1.00 (0.15)          0.55 (0.68)       0.67 (0.43)
                  Industry – Academe      0.008** (–1.32)      0.84 (–0.39)         0.83 (–0.39)      0.32 (–0.51)
                  Trade – Academe            0.96 (–0.30)      1.00 (–0.10)          0.86 (0.42)       1.00 (0.04)
                  Government – Consumer       1.00 (–0.27)     0.46 (–0.86)         0.52 (–0.80)      1.00 (–0.18)
                  Industry – Consumer     0.000*** (–2.04)   0.02* (–1.40)      0.000*** (–1.87)   0.006** (–1.11)

                  Trade – Consumer            0.29 (–1.01)     0.19 (–1.11)         0.21 (–1.06)      0.36 (–0.64)
                  Industry – Government   0.002** (–1.77)      0.74 (–0.53)         0.11 (–1.07)     0.03* (–0.93)
                  Trade – Government          0.59 (–0.74)     0.99 (–0.24)         0.98 (–0.26)      0.66 (–0.47)
                  Trade – Industry            –0.14 (1.03)      0.96 (0.29)          0.30 (0.81)       0.52 (0.47)
                  F(Sig.)                  6.87 (0.000***)     2.47 (0.05*)      5.12 (0.000***)    4.08 (0.004**)
* p < 0.05, ** p < 0.01, *** p < 0.001.
You can also read